ebruchez / darius-xml.js Goto Github PK

An XML parser based on Apache Xerces running in the Web browser via Scala.js.

See the demo for a quick feel.

Darius XML is able to parse a number of XML files already, but hasn't been tested to any serious extent.

There is not yet a usable API, whether for JavaScript or for Scala (issue #2).

If you need to process XML documents, you need an XML parser. Web browsers all embed one, and there is even a standard API for this, but:

• there is no support for any XML parsing or XML DOM within Web workers
• error handling is awkward and varies between browsers
• each browser has its own implementation, so parsing behavior might vary

There are a numbers of XML parsers for JavaScript, but as of April 2015 I have found none that can pretends to a minimum of compliance with the XML specification.

Since it doesn't seem wise to start writing an XML parser from scratch in 2015, I thought that starting with the time-tested Apache Xerces (which is also Java's built-in XML parser), which has more than 15 years of development behind it, might be a good idea.

But Xerces is written in Java. So in order to run it in the browser, I converted a subset of the code to Scala (first automatically, then fixing issues manually) in order to compile it with Scala.js.

The following are the immediate goals:

• provide XML parsing (including parsing the internal DTD subset if present)
• work in evergreen browsers
• have Scala and JavaScript APIs

The following are explicit non-goals at the moment:

• be a validating parser (whether with DTD or XML Schema)
• support SAX, DOM, JAXP APIs, XML Schema, XInclude, XPath, and XML 1.1
• remain API-compatible with Xerces
• support obsolete or failed features (XML without namespaces, XML 1.1)
• look like idiomatic Scala

Benefits:

• The parser reuses code which has been tested for more than 15 years.
• There is no need to write a parser from scratch. (You might think it is easy, being "just pointy brackets", but doing it properly, following the XML 1.0 and Namespaces in XML 1.0 specs, is in fact pretty hard.)

Drawbacks:

• As excellent as Scala.js is, the resulting library is probably larger than a library written from scratch in JavaScript or languages closer to it. In addition to the fact that Scala.js works with higher-level constructs than plain JavaScript, there can also be leftover bloat from the original Java library, and (at first at least) the Scala version depends on Scala collections (such as HashMap) which add some weight.
• The Scala code is the result of a translation from Java which means that errors might have been introduced. The best way to address this is to run a solid test suite on the Scala.js version.

The current demo app, including the XML parser:

• weighs 144 KB of compressed JavaScript
  • including relevant Scala library code
  • excluding jQuery, which is used for the demo app
• has 82 Scala files (including a few files for the demo)
• has 11,867 lines of non-blank, non-comments, non-test Scala code
  • impl: 7,627 lines
  • util: 2,630 lines
  • xni: 767 lines
  • parsers: 664 lines
  • demo: 139 lines
  • api: 40 lines

For comparison:

• a compressed Java JAR containing only the Xerces parser is 206 KB
• a compressed Java JAR of the full Xerces parser is 1.4 MB

TBD

sbt fastOptJs
sbt fullOptJs

Xerces is provided under the Apache 2 license. This means that the Scala files directly translated from Xerces are also released under that same Apache 2 license.

Files specific to Darius XML are also under the Apache 2 license.